Digitala Vetenskapliga Arkivet

Ändra sökning
RefereraExporteraLänk till posten
Permanent länk

Direktlänk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Inequality is Information: On its Quantification and Decomposition
Uppsala universitet, Teknisk-naturvetenskapliga vetenskapsområdet, Matematisk-datavetenskapliga sektionen, Institutionen för informationsteknologi, Avdelningen för datorteknik.
2025 (Engelska)Doktorsavhandling, monografi (Övrigt vetenskapligt)
Abstract [en]

Inequality and information are concepts from largely different communities. Information theory originated as a mathematical tool to study communication systems, while inequality theory evolved in economics and sociology. These areas developed measures to provide important quantitative insights for their respective applications. However, both domains also need the ability to decompose their measures to shed light on the structure of inequality within a population, or to understand how different pieces of information are provided and interact. 

In 2010, researchers in information theory developed a technique to decompose how information is provided redundantly by multiple sources, uniquely by a particular source, and synergistically by the interactions between sources. The proposed framework found favor in the community, but the decomposition measure was strongly criticized. Despite significant research efforts, no suitable replacement could be found. The primary contribution of this work is a solution to this open question: We developed a decomposition measure to non-negatively quantify the partial contributions of an arbitrary number of sources about a target with practical operational interpretation. 

Surprisingly, the underlying representation we used for our decomposing is equivalent to the representation for quantifying inequality. This identified relationship between measures of information and inequality enabled the direct transfer of results. Consequently, we generalized established inequality measures into a new family and provided a novel decomposition that characterizes inequality by the redundant, unique, and synergetic interactions between attributes of individuals. Finally, we demonstrate that subgroup decompositions from inequality theory highlight a recursive subgroup structure of information measures. 

The developed techniques can directly provide novel insights for studying applied questions in complex systems, information theory, economics, and sociology. Our tools enable professionals to gain deeper insights, understand the impact of changes and make informed decisions.

Ort, förlag, år, upplaga, sidor
Uppsala: Acta Universitatis Upsaliensis, 2025. , s. 166
Serie
Uppsala Dissertations from the Faculty of Science and Technology, ISSN 1104-2516 ; 157
Nyckelord [en]
Information Theory, Partial Information Decomposition, Inequality, Attribute Decomposition, Redundancy, Synergy
Nationell ämneskategori
Datavetenskap (datalogi)
Forskningsämne
Datavetenskap med inriktning mot datorkommunikation
Identifikatorer
URN: urn:nbn:se:uu:diva-544712ISBN: 978-91-513-2341-1 (tryckt)OAI: oai:DiVA.org:uu-544712DiVA, id: diva2:1919737
Disputation
2025-02-17, 101121 (Sonja Lyttkens), Ångström, Regementsvägen 1, Uppsala, 09:15 (Engelska)
Opponent
Handledare
Forskningsfinansiär
Myndigheten för samhällsskydd och beredskap, MSB , MSB 2018-12526Tillgänglig från: 2025-01-14 Skapad: 2024-12-09 Senast uppdaterad: 2025-01-16

Open Access i DiVA

UUThesis_Mages,T-2025(1404 kB)368 nedladdningar
Filinformation
Filnamn FULLTEXT01.pdfFilstorlek 1404 kBChecksumma SHA-512
1106c613abcfd5777378cdac48cb1032360b059d789d6eda2ded481f4b96785b6356da52b94ff137c1729b9c9c35c4380e5bd3d31689836311e50123701c4122
Typ fulltextMimetyp application/pdf
Köp publikationen >>

Sök vidare i DiVA

Av författaren/redaktören
Mages, Tobias
Av organisationen
Avdelningen för datorteknik
Datavetenskap (datalogi)

Sök vidare utanför DiVA

GoogleGoogle Scholar
Totalt: 370 nedladdningar
Antalet nedladdningar är summan av nedladdningar för alla fulltexter. Det kan inkludera t.ex tidigare versioner som nu inte längre är tillgängliga.

isbn
urn-nbn

Altmetricpoäng

isbn
urn-nbn
Totalt: 1344 träffar
RefereraExporteraLänk till posten
Permanent länk

Direktlänk
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Annat format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annat språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf